Edge flames were investigated in a methane/O 2 /N 2 counterflow diffusion flame burner. In a typical experiment, a stable counterflow diffusion flame in an axysymmetric configuration was perturbed by lowering the relevant Damkö hler number slightly below the extinction value, Da ext . As a result, the flame extinguished in the vicinity of the burner axis where conditions were uniform. An edge flame extinction front quickly propagated in the radial direction, turned into an ignition edge flame, and eventually stabilized as a standing triple flame at a radial position larger than the burner radius. This sequence of events resulted from an increase of Da as a function of the radial direction, consequent to a decrease in the strain rate in the radial direction. The edge flame propagation velocity in the ignition mode was measured for propagating edge flames at moderate Da and for standing triple flames at large Da, using a combination of laser Doppler velocimetry of seeded particles, formaldehyde planar laser-induced fluorescence, and natural chemiluminescence imaging. The propagation velocity, nondimensionalized with the premixed laminar flame speed of the unburned stoichiometric mixture, was correlated with Da. The latter was calculated using a thermal diffusive model and velocity measurements. The nondimensional velocity reached a value of 2.6 at large Da, in good agreement with the estimated square root of the ratio of the unburned gas density to the burned gas density, as suggested by scaling considerations.

Using a combination of HCHO planar laser-induced fluorescence and laser Doppler velocimetry measurements, the extinction behavior of methanol counterflow diffusion flames was examined experimentally under conditions in which the extinction was brought about by a vortex generated on the oxidizer side. Comparisons were made with quasi-steady extinction results for the same flames. It was found that the flames can withstand instantaneous strain rates as much as two-and-a-half times larger than the quasisteady ones. The finding was rationalized phenomenologically by comparing the characteristic times of the problem, that is, the mechanical time, the chemical time, and the vortex turnover time. Specifically, estimates of these times yielded the following ordering: s ch Ͻ s vort Ͻ s m . As a result, the vortex introduced an unsteady effect in the outer diffusive-convective layer of the flame, while the inner reactive-diffusive layer behaved in a quasi-steady manner. Consequently, the flame was subject to a damped strain rate through the outer layer. Results from a simple analytical model showed that the difference between vortexinduced extinction and quasi-steady extinction was much more modest in terms of instantaneous scalar dissipation rate or Damkö hler number. Furthermore, the temporal history of the strain rate was found to be necessary to determine the effective strain rate felt by the flame. Implications of these findings for turbulent diffusion flame modeling by the flamelet approach are discussed.

No abstract

The extinction behavior of methanol counterflow spray diffusion flames was investigated using a combination of formaldehyde planar laser-induced fluorescence (PLIF) and phase Doppler measurements. Extinction was brought about quasi-steadily, by progressively increasing the flow rates of both oxidizer and fuel side, and unsteadily, by generating a vortex on the oxidizer side. The unsteady experiments yielded values of extinction strain rates a factor of 2 larger than the quasi-steady values. The greater robustness of the spray flame under unsteady perturbation was explained phenomenologically by estimating the timescales involved in the process. It was found that the vortex introduces unsteady effects in the outer diffusiveconvective layer of the flame. The inner reactive-diffusive layer, on the other hand, behaves in a quasisteady manner, since the characteristic chemical time is much smaller than the characteristic unsteady time. As a result, even though the instantaneous strain rate is much larger than the quasi-steady extinction strain rate, the flame is subject to a damped strain rate through the outer layer. An estimate of the thickness of the mixing layer, based on formaldehyde PLIF, provided a convenient means to compare the scalar dissipation rate and the Damkö hler number between the two extinction modes, bypassing the need for detailed species measurements for the assessment of the mixture fraction and its gradient. Such a comparison showed that the difference between the two extinction modes was reduced to 25% on the average, consistent with expectations based on flame structure models from asymptotic theory. Spray flames exhibited longer time delays between the onset of extinction and reignition, as compared to gaseous flames. Estimates of the relevant Stokes number suggested that the difference may be attributed to droplet inertia effects.

The interaction of laminar vortices with a methanol spray counterflow diffusion flame was studied experimentally with vortices generated from either the fuel side or the oxidizer side. The overall stoichiometry was such that the flame resided on the fuel side of the gas-stagnation plane. Local extinction and subsequent reignition were investigated as the circulation of the vortex was varied. It was found that extinction required vortices of larger circulation if generated from the oxidizer side. The effect was attributed to stretching, and possibly, partial dissipation of the vortex, as it approached the stagnation plane before interacting with the flame. The robustness of the spray flame to vortex-induced extinction was compared with that of a similar gaseous flame. The spray flame was found to be comparatively weaker. Of the two potential culprits for such a difference, namely, the energetic handicap of spray flames due to the latent heat of vaporization of the liquid fuel and droplet inertia, the former was found to be the dominant factor. After extinction occurred, a hole was created in the diffusion flame, confining the combustion process to an annular region. The flame was then able to propagate back toward the centerline, re-establishing itself as a flat diffusion flame. The time interval for this reignition process was investigated as a function of the vortex circulation. It was found that, if the vortex approached the flame from the fuel side, the reignition time was much shorter than when the vortex was injected from the oxidizer side, and decreased for increasing values of the vortex circulation. In contrast, the reignition time increased with the circulation, if the vortex approached the flame from the oxidizer side. Only in the latter case, droplet inertia played a role in the reignition process.